Image Classification on CIFAR-10 (test) (Pruning Error Analysis)
5.77Error after PruningBi-CNN-Pruning
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Bi-CNN-PruningCNN architecture=ResNet562026.04 | 5.77 | 6.9 | -1.13 | 23.21 | |
| Bi-CNN-PruningCNN architecture=ResNet1102026.04 | 6.41 | 6.2 | -0.21 | 26.33 | |
| Greedy-FilterCNN architecture=ResNet1102026.04 | 6.7 | 6.47 | -0.23 | 38.63 | |
| Greedy-FilterCNN architecture=ResNet562026.04 | 6.94 | 6.96 | -0.02 | 27.67 | |
| DeepPruningESCNN architecture=ResNet1102026.04 | 7.42 | 6.2 | 1.22 | 16.74 | |
| WIEACNN architecture=ResNet1102026.04 | 7.69 | 7.47 | 0.22 | 46.23 | |
| WIEACNN architecture=ResNet562026.04 | 7.82 | 6.72 | 1.1 | 20.84 | |
| Two-Phase PruningCNN architecture=ResNet1102026.04 | 7.89 | 6.82 | 1.07 | 40.89 | |
| DeepPruningESCNN architecture=ResNet562026.04 | 8.12 | 6.63 | 1.49 | 21.3 | |
| Auto-Balanced FilterCNN architecture=ResNet1102026.04 | 8.27 | — | — | 40 | |
| Auto-Balanced FilterCNN architecture=ResNet562026.04 | 8.56 | 6.96 | 1.6 | 41.51 | |
| Two-Phase PruningCNN architecture=ResNet562026.04 | 8.56 | 7.33 | 1.23 | 50.9 |